Page 542 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Use Case 12: AI-Based System for Assessing Health and Moisture

Content of Concrete

Organization: Mbeya University of Science and Technology

Country: Tanzania
Godfrey Samwel Butete, godsamwel4@ gmail .com, +255757315443

1 Use Case Summary Table

Item Details
Category Manufacturing

Problem Addressed Manual inspection of concrete structures is time-consuming, often
destructive, and lacks real-time monitoring capability.

Key Aspects of Solution AI-based non-destructive evaluation using Convolutional Neural
Networks (CNNs) for crack detection and sensor data for moisture
analysis.

Technology Keywords AI, CNN, Regression, IoT Sensors, Edge Computing,
Structural Health Monitoring
Data Availability Combination of public datasets (SDNET2018 dataset [3]) and private
data collected at MUST labs.
Metadata (Type of Data) Visual data (images of cracks), sensor data (moisture, strain, load)

Model Training and CNNs trained with image datasets; regression models trained on
Fine-Tuning sensor logs; uses transfer learning.
Testbeds or Pilot University lab environment, local construction firms (planned)
Deployments

Code repositories Not yet hosted in Github

506